作物遥感估产方法对比及一致性评价

Comparison and consistency analysis of crop remote sensing methods for yield estimation

关键词:  作物遥感估产  对比实验  一致性分析  遥感经验模型

Key words:  crop remote sensing estimation  comparative experiments  consistency analysis  remote sensing empirical model

摘要:

【目的】 针对不同作物遥感估产方法结果存在的差异性，对多情景下遥感估产方法进行对比及一致性分析，以探讨造成作物估产结果差异的可能因素。【方法】 文章按照“数据+模型”的组合方式对遥感估产方法进行归类，随后在不同年份、地域和作物之间进行估产实验，并对估产结果和一致性进行分析，以比较组合方法在多情景下的作物估产效果。【结果】 （1）环境数据和多源遥感数据均能提高估产结果的准确性，统计回归方法比机器学习方法的稳定性更高；（2）不同方法对地形和环境因素的处理方式差异降低了估产结果的一致性，距水体的距离是影响作物产量的因素之一。【结论】 数据源和模型的选择能显著影响作物估产效果；估产结果的一致性分析可以更好地识别不同数据源和模型对空间分布的影响。

Abstract:

［Purpose］ This study compares several commonly used methods for remote sensing based crop yield estimation across multiple scenarios to evaluate their consistency and variability，aiming to identify the possible factors contributing to these differences.［Method］ Remote sensing yield estimation experiments were initially grouped according to the combination of 'data + model' and then conducted across multiple years，regions，and crop types.The results from different methods were compared and analyzed for consistency，with a primary focus on evaluating their performance in multi-scenario crop yield estimation.［Result］ （1）The inclusion of environmental data，together with multi-source remote sensing data，significantly enhanced the accuracy of yield estimation models. Statistical regression methods showed greater stability compared to machine learning methods.（2）Inconsistencies in the treatment of topography and environmental factors between methods reduced the consistency of yield estimates.Furthermore，proximity to water bodies was a key factor influencing crop yield variability.［Conclusion］ Selecting appropriate data sources and models is essential for achieving accurate crop yield estimation.In addition，the consistency analysis provides valuable insights into how variations in data sources and model choices influence the spatial distribution of crop yields.